Fuel 82 (2003) 1451–1463

www.fuelfirst.com

On the optimization of boiler efficiency using bagasse as fuelq

Jorge Barrosoa, Félix Barrerasb,*, Hippolyte Amavedac, Antonio Lozanob
a
CECYEN, Universidad de Matanzas, Autopista a Varadero, 44740 Matanzas, Cuba
b
LITEC/CSIC, Maria de Luna 10, 50015 Zaragoza, Spain
c
Fluid Mechanics Group, University of Zaragoza, Maria de Luna 3, 50015 Zaragoza, Spain
Received 30 November 2002; revised 31 January 2003; accepted 16 February 2003; available online 20 March 2003

Abstract
The present investigation has been carried out in order to increase the efficiency of the RETAL-type boiler, used in the Cuban sugar mills.
Test methods generally used in the evaluation process and further adjustment of the boilers operation have been analyzed, pointing the
attention on the importance of the stoichiometric ratio and steam power on the overall efficiency. Important general rules have been extracted
from the complete regular tests following ASME and GOST methodologies, and, as a result, a simplified test code has been obtained. Boiler
design optimization has also been achieved determining the optimum waste heat recovery scheme from both, thermal and economical
viewpoints. As a result, the optimal stack gas temperature has been calculated as well as the range of the optimal value for the excess air
fraction. Their influence on the efficiency has been analyzed and the total costs determined. Once the total costs are included in the analysis,
the most efficient low-temperature heat recovery scheme results to be the combination of an economizer followed, in the direction of the
exhaust gas flow, by an air heater.
q 2003 Elsevier Science Ltd. All rights reserved.
Keywords: Biomass; Bagasse-boilers; Efficiency; Optimization

1. Introduction efficiency boilers. The designs referred to as ‘fuel cell’ and

Bagasse is the matted cellulose fiber residue from sugar
cane that has been processed in a sugar mill. Previously,
bagasse was burned as a means of solid waste disposal.
However, as the cost of fuel oil, natural gas, and electricity
increased after the energy crisis in 1970, special attention
was paid to alternative fuels in an efficient way. Conse-
quently, conception of bagasse combustion changed and it
has come to be regarded as a biomass fuel rather than refuse.
Another important aspect is the increasing demand of
bagasse as raw material for paper, furniture, and other
industries. For all these reasons, the saving of this product
has become one of the main objectives of the Cuban sugar
cane industry. The actual tendency is to use bagasse as fuel,
especially for co-generation of electric power and steam, to
increase its contribution to the country’s energy supply.
Cuba has, in bagasse, a renewable energy source
representing 30% of its total energy consumption. Tra-
ditionally, Cuban sugar mills were equipped with low
* Corresponding author. Tel.: þ 34-976-716-303; fax: þ34-976-716-456.
E-mail address: felix@litec.csic.es (F. Barreras).
q
Published first on the web via Fuelfirst.com—http://www.fuelfirst.com
0016-2361/03/$ - see front matter q 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0016-2361(03)00061-9
‘horseshoe’ boilers were those typically used for bagasse
combustion. In these boilers, bagasse is gravity-fed through
chutes and burned as a pile. Nowadays, bagasse is burned in
spreader stoker boilers, replacing the combustors that use
pile type approaches, and improving combustion efficiency.
Furthermore, the use of additional heat transfer surfaces, as
air heaters, economizers, etc. allows for a reduction of the
stack temperature below 200 8C. With these improvements,
efficiency of the boilers can be increased up to 70%.
This experimental research, developed by the Center of
Studies on Combustion and Energy (CECYEN) of the
University of Matanzas, has been devoted to optimize both
the operation and design parameters of RETAL boilers
using bagasse as fuel in order to increase the efficiency of
Cuban sugar mills. Special attention has been paid to the
optimization of stoichimetric ratio as well as the stack gases
temperature, for their influence on the principal heat losses
and, consequently, on the overall efficiency of the boiler.
2. Boiler and fuel characteristics
Sugar cane is a large grass with a stalk that grows
2 –5 m tall. Only the stalk contains sufficient sucrose to be
1452 J. Barroso et al. / Fuel 82 (2003) 1451–1463

Nomenclature g heat transfer coefficient (kW/(m2 K))

h overall boiler efficiency (%)
a ash ratio respect to the total ash in the bagasse
m molecular weight
(kgash in refuse/kgash in fuel)
C thermal efficiency
A ash content in fuel or in refuse (%)
B bagasse consumption (kg/s)
Bef equivalent fuel-oil consumption (kg/s)
C carbon content in bagasse (%) Subscripts
Cuf unburned carbon in the fly ash (kgc/kgfa)
CO carbon monoxide in stack gas (%) ah ash hopper
CO2 carbon dioxide in stack gas (%) AH air heater
D steam power (t/h) b boiler
DHC heat of combustion (kJ/kg) ba bottom ash
F heat transfer area (m2) bd bagasse dryer
G refuse collected per time (kgrefuse/s) ea external air
H hydrogen content in bagasse (%) ec economizer
I enthalpy (kJ/kg) ef equivalent to fuel oil
ma=f air-to-fuel mole number ratio (moleair/molefuel) eg exhaust gas (stack)
O oxygen content in bagasse (%) f furnace
O2 oxygen percentage in stack gas (%) fa fly ash
q2 exhaust gas loss (%) gt generating tubes
q3 chemical carbon loss (%) H chemical composition given by Hugot
q4 fixed carbon loss (%) i different heat losses, heat transfer surfaces or
q5 conduction heat loss (%) refuse collected
QPl bagasse heating value (as received) (kJ/kg) L laboratory analysis
QPef commercial fuel oil heating value (kJ/kg) sh superheater
P annual cost per m2 of surface area ($/(m2 yr)) w water-walls
Pef equivalent fuel-oil cost ($ s/(kg yr))
RCO=f kg of CO produced per kg of fuel
S sulfur content in bagasse (%)
Superscripts
T temperature (8C)
W bagasse moisture (%)
e entrance
Z total cost ($/yr)
o theoretical
p ultimate bagasse analysis (as received)
nom nominal
Greek symbols

a stoichiometric ratio

processed into sugar. All other parts of the sugar cane
including leaves, roots, etc. are termed ‘trash’, which
should be eliminated through the harvesting process. Once
inside the mill, juice is extracted in the plant milling
section by passing the chopped and crushed cane through
a series of grooved rolls. The cane remaining after milling
is the bagasse. Usually, it is a biomass-type fuel of
varying composition, consistency and heating value.
These characteristics depend on the climate, type of soil
where the cane is grown, cane type, harvesting method,
amount of cane washing, and efficiency of the milling
plant.
The experiments were carried out in three RETAL
boilers of two different Cuban sugar mills. Fig. 1 shows a
detailed sketch of the main thermal surfaces of these
facilities. The total height and depth of the boiler are 10.6
and 10.92 m, respectively, and the width (not shown in
the figure) is 8 m. Summarizing the main characteristics, a
nominal steam power of 45 t/h is achieved for an
approximate bagasse consumption of 22 t/h; with a
pressure and temperature of the superheated steam of
1.9 MPa and 320 8C, respectively. Bagasse fed to these
boilers enters the furnace through five fuel chutes and is
spread mechanically. The major part of the bagasse
characterized by small and light pieces, burns in
suspension. Simultaneously, large pieces of fuel are
spread in a thin even bed on a stationary grate. An
average ultimate (dry) analysis of the fuel used in the tests
gave a 46.27% (in weight) of carbon, 6.4% of hydrogen,
43.33% of oxygen, 0% of nitrogen, 0% of sulfur, and 4%
of ash. The moisture content of the bagasse ranged from
48 to 52% for all the analyzed samples.
 J. Barroso et al. / Fuel 82 (2003) 1451–1463
Fig. 1. Sketch of a RETAL bagasse-boiler showing the principal thermal surfaces (numbers) and location of measuring points: (1) furnace grill, (2) spreader stoker, (3) furnace, (4) superheater, (5) drums, (6)
generating tubes, (7) air heater, (8) economizer, (9) exhaust gases duct, (10) air supply fan, (11) air extraction fan, (12) smokestack and (13) the ash hopper in the U-turn of the exhaust gas duct. Letters refer to the
measured parameters; A; ash concentration; GA, exhaust gas composition analysis; N; motor power; P; pressure; r; revolutions per minute; R; residual weight; T; temperature, and W; bagasse moisture
percentage.

1453
1454 J. Barroso et al. / Fuel 82 (2003) 1451–1463
3. Operational test procedure
More than 60 tests were performed, attending the ASME
[1] and GOST [2] recommendations for solid and liquid
fuels. Each test comprised three stages, namely: preparation,
measurements, and laboratory analysis. According to the
standard procedures, one should wait at least 24 h after start-
up of the boiler and 2 h after cleaning of the bottom ash, the
ash hopper located in the U-turn of the flue gas duct (see
number 13 in Fig. 1), and the heat transfer surfaces before
starting a test. The boiler should reach, and maintain, a
steady state for at least 8 h before starting the test. The fuel
chute and the stationary grate must also be cleaned 1 h
before starting, and the speed of rotation of the spreader
stokers fixed. Some trays have to be placed in the proper
locations for refuse collection, and the fly ash wet scrubber
has to be cleaned as well.
Boiler measurements and laboratory analyses are per-
formed along the following 9 hours. The first four hours are
dedicated to measure all the boiler parameters every
15 minutes. Every half hour, stack gas composition (O2,
CO, and CO2) is determined and bagasse samples collected
for the determination of their moisture and ash contents. The
furnace temperature is also measured every 15 min using
water-cooled suction pyrometers. The sampling and
measurement locations are shown in Fig. 1.
Laboratory work begins with refuse collection from all
the different locations (ash bottom, ash hopper and web
scrubber). In the following five hours, moisture and ash
contents of the bagasse and solid samples from the fly ash
hoppers, wet scrubber, and bottom ash hopper are analyzed.
When all the data are assembled, a statistical analysis
determines the mean and standard deviation for each
parameter. If a steady state has not been achieved in the
boiler, the test must be rejected.
As it is well-known, the overall efficiency of a boiler can
be calculated using both direct and indirect methodologies.
The direct measurement of the bagasse consumption, B; is
always subjected to many error sources. For this reason, in
the present study, efficiency has been calculated using the
indirect methodology. In general, this method relates the
efficiency ðhÞ of the boiler with the different heat losses
through the equation
X fly ash, Gfa ; ash hopper, Gah ; and bottom ash, Gba ;
h ð%Þ ¼ 100 2 qi ð1Þ
P
where qi ¼ q2 þ q3 þ q4 þ q5 : In this equation, q2
represents the exhaust gases heat loss, q3 and q4 are
the chemical and fixed carbon loss, respectively, and q5
the conduction heat loss from the external walls of the
boiler. To quantify the heat losses, the following
equations are used [2]:
!
100 2 q4
q2 ð%Þ ¼ ðIeg 2 ab Iea Þ ð2Þ
QPl
DHCCO
q3 ð%Þ ¼ RCO=f £ 100 ð3Þ
QPl
 
DHCC AP X ð100 2 Ai Þ
q4 ð%Þ ¼ ai ð4Þ
QPl Ai
P
gT Fi ðti 2 tea Þ
q5 ð%Þ ¼ ð5Þ
BQPl
Here, Ieg and Iea are the exhaust gases and external air
enthalpy, respectively, ab the stoichiometric ratio at the
exit of the boiler, QPl the bagasse heating value (as
received), DHCC the carbon heat of combustion, DHCCO the
CO heat of combustion, AP the ash contents of bagasse
from ultimate analysis (as received) and RCO=f is the rate
of kilograms of CO produced during the combustion of
one kilogram of fuel.
The stoichiometric ratio, a ¼ ma=f =moa=f ; is defined as the
ratio of the actual air-to-fuel mole number ratio ðma=f Þ to the
theoretical one ðmoa=f Þ for the same experimental conditions.
In turn, the actual air-to-fuel mole number ratio ðma=f Þ is
defined as the theoretical number of moles of air plus the
extra moles due to excess air needed to achieve the complete
combustion of one mole of bagasse (in moleair/molefuel). As
it will be thoroughly discussed in Section 5.2, in order to
more accurately reproduce the physical influence of the
different parameters and heat losses in the statistical models,
two stoichiometric ratios have been defined, namely:
stoichiometric ratio at the furnace, af ; and stoichiometric
ratio at the exit of the boiler, ab : It should be pointed out that
in the case of ab ; the amount of surrounding air in-leakage
into the boiler due to non-air tightness, Da; is also included
in the total air mole number.
Ai refers to Afa ; Aah and Aba ; which correspond to ash
percentages in the fly ash, ash hopper and bottom ash,
respectively, obtained through laboratory analysis combust-
ing and weighting the different samples of refuse collected in
a special oven following the methodology of ASME [1] and
GOST [2]. In the same way, ai refers to the ratios of ash in
the fly ash, afa ; ash hoppers, aah ; and bottom ash, aba with
respect to the total ash in the fuel, in kgash in refuse/kgash in fuel.
From a mass balance of ash in the boiler, considering Gi as
the refuse collected per time unit in the different locations in
respectively, in kgrefuse/s, the following equation can be
written
BAP ðkgash =sÞ ¼ Gfa Afa þ Gah Aah þ Gba Aba ð6Þ
the different ash ratios ai ; in fly ash, afa ; ash hoppers, aah ; and
bottom ash, aba ; are defined by
Gi Ai
ai ðkgash in refuse =kgash in fuel Þ ¼ ð7Þ
BAP
and hence
1 ðkgash in refuse =kgash in fuel Þ ¼ afa þ aah þ aba ð8Þ
 J. Barroso et al. / Fuel 82 (2003) 1451–1463
To calculate the conduction heat loss, q5 ; from the external
wall to the surrounding area, the total heat lost (including
radiation) has to be considered. In Eq. (5), lT is the total heat
transfer coefficient in kW/(m2 8C), F is the total heat transfer
area of the external wall in m2, ti and tea are the external wall
and external air temperature (K), respectively. It has to be
noted that in Eq. (5), the fuel consumption, B; has to be
included for dimensional homogeneity. The unknown fuel
consumption in Eqs. (4) and (5) is calculated by an iterative
procedure.
4. Waste heat recovery scheme
The RETAL boiler was adapted from a German model
that used pulverized coal as fuel. Over the years, the boiler
has been redesigned, mostly by modifying its combustion
systems according to the changes in fuel type. However, a
careful study to optimize the waste heat recovery scheme
has never been specifically attempted. It is for this reason
that one of the aims of the present investigation was the
determination of the most suitable combination of low-
temperature heat transfer surfaces, from both thermal and
economic viewpoints, to obtain the optimum waste heat
recovery scheme.
Special attention has been paid to heat losses respect
to the exhaust gas, because they can reach up to 30% of
the total energy in the fuel. To obtain the optimal value
for the exhaust gases temperature, it is necessary to use
additional heat transfer surfaces such as an economizer,
air heater, bagasse dryer, or some combination of them.
However, the addition of new elements increases the
investment and operating costs of the boiler and hence,
the importance of establishing the optimal stack
temperature.
To solve this problem, a minimum total cost ðZÞ has to be
found through the equation
X placement order following the flue gas flow direction. The
Z ð$=yrÞ ¼ Pi Fi þ Pef Bef ð9Þ
where i is the type of recuperative heat transfer surface
(furnace water-walls, superheater, generating tubes, air
heater, economizer, and bagasse dryer); Pi the annual cost of
1 m2 of the surface i ($/(m2 yr)); Fi the heat transfer area of
surface i (m2), Pef the equivalent fuel-oil cost ($ s/(yr kg))
and Bef the equivalent fuel-oil consumption (kg/s). This
fuel-oil equivalence means the amount of commercial fuel-
oil with an average heating power ðQPef Þ of 41,868 kJ/kg
(and its price at the oil market), needed to yield the same
energy as the total bagasse consumed to produce a given
steam power. Once the efficiency is determined using the
indirect method previously described in Section 3, the total
bagasse consumption, B; is calculated by
e efficiency of a boiler, it was carefully determined using a
Dsh ðIsh 2 Iec Þ
B ðkg=sÞ ¼ P
£ 100 ð10Þ
ðQl hÞ
1455
e
where Dsh is the measured steam power in t/h and Ish and Iec
are the superheated steam and the fed water enthalpy,
respectively. The equivalent fuel-oil consumption Bef is
determined by
BQPl
Bef ðkg=sÞ ¼ ð11Þ
QPef
Using the common methodology to calculate the minimum
value of a function, the equation obtained to determine the
optimal stack temperature, considering all the heat transfer
surfaces is
dZ P dB dðPw Fw Þ dðPsh Fsh Þ dðPgt Fgt Þ
¼ ef ef þ þ þ
dTeg dTeg dTeg dTeg dTeg
dðPAH FAH Þ dðPec Fec Þ dðPbd Vbd Þ
þ þ þ ¼0
dTeg dTeg dTeg
ð12Þ
In this equation, Teg is the stack temperature; P and F define,
respectively, the cost and area for all thermal surfaces
considered. Subscript w indicates furnace water-walls; sh
superheater, gt generating tubes; AH the air heater; ec the
economizer, and bd the bagasse dryer.
Similar equations can be derived for hot air temperature
and bagasse moisture optimization. The computer code
developed for the optimization procedure is capable of
performing the simultaneous optimization of the stack and
hot air temperatures. At the same time, if a bagasse dryer is
considered in the waste heat recovery scheme, the bagasse
moisture can also be optimized. Obviously, as the available
remaining heat has to be transferred to the water, as a
consequence, an optimum water temperature is obtained as
well. The equations have to be adapted according to the
particular heat recovery scheme to be optimized. Five
combinations of retrofitted heat exchangers have been
considered in the present study. In naming the various
configurations, the different surfaces are listed in their
examined cases are: air heater – economizer –air heater (case
I), air heater – economizer (case II), economizer – air heater
(case III), economizer (case IV), and air heater –economi-
zer – bagasse dryer (case V). In each case, the name of the
first surface listed corresponds to that exposed to the highest
gas temperature.
5. Results and discussion
5.1. Bagasse heating value determination
In general, bagasse has a broad range of heating values,
extending from 6500 to 9150 kJ/kg (as received). Due to the
importance of this parameter in the determination of the
calorimeter on more than 1000 samples collected during
the tests. Results yielded an average heating value for
1456 J. Barroso et al. / Fuel 82 (2003) 1451–1463
the bagasse of 7738 ^ 100 kJ/kg, as received. However, in
most sugar mills, it is not possible to carry out such a
determination in their laboratories. An alternative method
has been considered, calculating the heating value using the
well-known equation [2]
QPl ðkJ=kgÞ ¼ 339:15CP þ 1256:10H P 2 108:86OP
2 25:12ðH P þ WL Þ ð13Þ
taking into account the general chemical composition given
by Hugot [3], but considering bagasse moisture and ash
contents from the samples measured in the laboratory tests.
As an example, the following relation is used to modify the
carbon composition
CP CHP
¼ confidence but keeping them as simple as possible, in
100 2 WL 2 APL 100 2 WH 2 APH
where C P is the carbon content in the bagasse, and W and A
are the bagasse moisture and ash contents. Subscripts H and
L indicate values given by Hugot and those experimentally
determined in the laboratory analysis, respectively. Similar
equations can be written for the other chemical components,
i.e. hydrogen and oxygen.
It was found that the lower heating values calculated
using Eq. (13) differed from experimental measurements by
less than 1%. The fact that experiments were performed in
two different sugar mills, having widely different methods
of harvesting and cane varieties, indicates that Eq. (13)
provides a means for quickly determining the heating value
of bagasse (as received) with a high confidence level, good
accuracy, and avoiding the more difficult and time-
consuming experimental measurements. In contrast to the
ultimate fuel analysis, bagasse moisture and ash contents are
relatively easy to measure and are accessible to every sugar
mill.
5.2. Optimization of the boiler operation
With the present experimental investigation, an adequate
methodology to determine the efficiency of bagasse-boilers
was established, adapting the ASME and GOST test code
evaluations to this particular fuel and type of boiler. The
principal drawback of the ASME and GOST methods is the
large time required and the high cost, including personnel,
of each test, as can be inferred from the description in
Section 3. For this reason, a special effort has been devoted
to optimize both the test itself and the boiler operation.
Important general rules have been extracted from the
complete regular tests, and, as a result, a simplified
Bagasse-boilers Industrial Test Code [4] has been elabo-
rated. In short, the stationary regime should be reached only
one hour before, and during the whole test, allowing
maximum fluctuations of 15% for the prescribed steam
power and its temperature, 3% for the water temperature
and 7% for the steam pressure and the stoichiometric ratio.
The determination of the bagasse moisture and ash contents
need to be performed only once during the test. The analysis
of exhaust gas composition is measured at the beginning and
at the end of the test. Validity of the simplified test code has
been demonstrated in more than 30 boilers.
To carry out the boiler optimization for different
operational regimes, experimental measurements have
been obtained from the full tests according to the ASME
[1] and GOST [2] procedure. Special attention was devoted
to obtain general charts relating measured parameters, such
as the stoichiometric ratio, steam power, etc. to the overall
boiler efficiency ðhÞ: As a consequence of the experimental
results, some important simplifications on both the fixed
carbon loss, q4 ; and conduction heat loss, q5 ; are considered.
At the same time, attention has also been focused to obtain
the needed statistical models, with a high level of
order to ease the efficiency evaluation of the boilers by
engineers at the sugar mill factory.
Determination of the conduction heat loss, q5 ; in an exact
way is quite difficult, requiring the measurement of all
external wall temperatures as well as the determination of
the heat transfer coefficient as commented in Section 3.
From the experimental tests, it is concluded that q5 shows
only a strong dependence on steam power. For this reason, a
simplified equation relating q5 with the steam power
commonly used in this type of boilers [2] was considered,
namely
 nom sffiffiffiffiffiffiffiffi
Dsh 100
q5 ð%Þ ¼ ð14Þ
Dsh Dnom
sh
Results obtained using both Eqs. (5) and (14) demonstrated
a very good agreement between the complete thermal
analysis and the simplified one. It is for this reason the Eq.
(14) was included in the recommended simplified Bagasse-
boiler Industrial Test Code, and also included in the
computer code used to optimize the boiler operational
regime, with the only measurement of the steam power to
determine this heat loss.
Considering the physical influence of the fixed carbon
loss ðq4 Þ on the remaining heat losses (q2 and q3 ), it must be
the first of all the heat losses to be evaluated in the efficiency
calculation. For the terms inside the brackets in Eq. (4),
experimental measurements during the tests performed
demonstrated the validity of the following inequality
ð100 2 Afa Þ ð100 2 Aah Þ ð100 2 Aba Þ
afa s aah þ aba ð15Þ
Afa Aah Aba
which means that the terms corresponding to ash hopper
and bottom ash can be neglected when compared to the
fly ash one.
The terms ð100 2 Ai Þ in Eqs. (4) and (15) are, by
definition, the unburned fuel (carbon) for the refuse
collected in the different locations. Having in mind that q4
is expressed as an unburned loss, it is convenient at this
 J. Barroso et al. / Fuel 82 (2003) 1451–1463
moment to introduce the relation Cuf ¼ 100 2 Afa as the
unburned carbon in the fly ash. For this reason, Eq. (4) can
be rewritten as
 
DHCC AP Cuf
q4 ð%Þ ¼ ð16Þ
QPl 100 2 Cuf
In this equation, afa has been considered equal to 1 (Eq. (8))
taking into account the experimental results and calcu-
lations, with a high confidence level.
It can be observed that the above simplifications to
calculate for q4 and q5 ; reduce and simplify the number of
statistical models to be fitted to obtain the claimed
efficiency chart. In the next paragraph, the statistical
models fitted to calculate Cuf ; CO/(CO þ CO2) and stack
temperature, Teg ; will be discussed. It should be noted that
the last parameter, Teg ; is needed to calculate the exhaust
gases enthalpy in Eq. (2).
In order to obtain a simplified model, the influence of the
stoichiometric ratio in the furnace ðaf Þ and steam power
ðDsh Þ on the unburned carbon in the fly ash, Cuf ; has been
plotted. Results of the experimental measurements are
depicted in Fig. 2. As can be seen, the unburned carbon
increases with increasing steam power and decreases with
increasing stoichiometric ratio in the furnace. As steam
power is raised at a constant stoichiometric ratio, both the
amount of bagasse fed and the combustion air flow rate
increase, since the air volumetric flow rate per unit weight of
bagasse is fixed. This, in turn, increases the average gas
velocity in the furnace and the fraction of fuel that burns in
suspension, rather than in the bed on the stationary grate.
The shorter residence time available for combustion in
suspension results in an increased unburned carbon carry-
over and poorer combustion performance. Therefore,
Fig. 2. Performance of the unburned carbon in fly ash vs. stoichiometric ratio at the exit of the furnace ðaf Þ for different values of steam power. The depicted
solid lines correspond to the different steam powers given by Eq. (17).
1457
when steam power and bagasse consumption are increased,
a higher stoichiometric ratio in the furnace is needed to
achieve the same carbon conversion ðCuf Þ: Taking into
account all the experimental data, a statistical model is fitted
Cuf ðkgc =kgfa Þ ¼ 0:854965 þ 0:002724Dsh
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1:3416
2 0:592243 af þ pffiffiffiffiffi ð17Þ
Dsh
reproducing the experimental dependence of Cuf on both the
stoichiometric ratio at the furnace, af ; and the steam power,
Dsh : In this equation Dsh has units of t/h.
Once q4 is calculated, q3 and q2 can also be evaluated. To
determine the chemical carbon heat loss ðq3 Þ; the parameter
RCO=f needs to be calculated by the equation
!  
C P þ 0:375SP CO 100 2 q4
RCO=f ¼ mCO=C
100 CO þ CO2 100
mCO=C being the CO-to-C molecular weight ratio. C P and SP
are the carbon and sulfur contents of bagasse from ultimate
analysis (as received). CO and CO2 are the carbon
monoxide and dioxide concentration in the stack gases,
respectively. Substituting RCO=f in Eq. (3), only the term
containing CO and CO2 remains to be determined from the
experiments in order to establish a correlation for the
chemical carbon loss, q3 : Results are presented in Fig. 3,
showing a linearly decreasing dependence of this term for
increasing stoichiometric ratio, and a weak influence of the
steam power. For clarity, in this plot, error bars are only
displayed for the measurements corresponding to steam
powers of 20 and 50 t/h, respectively. The remaining
experimental points represent the mean values for
1458 J. Barroso et al. / Fuel 82 (2003) 1451–1463

Fig. 3. Behavior of CO and CO2 concentration in the stack gases as a function of stoichiometric ratio at the exit of the furnace ðaf Þ and steam power. The fitted
lines correspond to Eq. (18) for values of the steam power of 20 and 50 t/h (lower and upper lines).

the different stoichiometric ratios and steam powers previously determined for all boilers tested and yielded a
analyzed. The statistical model relating carbon monoxide, roughly constant value of 0.2.
expressed as a fraction of the total carbon oxides, to both af Finally, to calculate the exhaust gases heat loss, q2 ; the
and Dsh ; is exhaust gas enthalpy, Ib ; needs to be known. This enthalpy
CO 0:0165 depends on the stack temperature, Teg : The experimental
¼ 0:0275 2 0:01485af 2 pffiffiffiffiffi ð18Þ data show a linear dependence of Teg on the stoichiometric
CO þ CO2 Dsh
ratio, as seen in Fig. 4. As in Fig. 3, errors bars are only
It is important to note that the physical parameters measured depicted for the lowest and highest values of the steam
in the tests are the O2, CO and CO2 concentration in the power (20 and 50 t/h). Fitting a curve to the measurements
exhaust gases, which are used to calculate the stoichiometric relating Teg to af and Dsh ; the following equation is
ratio, ab ; at the exit of the boiler applying the equation [2] obtained
1 43:22
ab ¼   ð19Þ Teg ð8CÞ ¼ 172:32 þ 24:76af þ pffiffiffiffiffi 2 0:213ðDsh Þ0:33
O2 2 0:5CO Dsh
1 2 3:76
100 2 O2 2 CO2 2 CO ð22Þ
where O2 ; CO and CO2 are the stack gases composition as shown in Fig. 4 by the solid lines for 20 and 50 t/h,
analysis. This equation is obtained using the combustion respectively. Even when the dependence of Teg on Dsh is
reactions as a function of the mole number, following the weak, it is noteworthy that the stack temperature is raised
definition of the stoichiometric ratio previously stated in as the steam power decrease. When the steam power is
Section 3. decreased for a constant stoichiometric ratio at the furnace,
However, Eqs. (17) and (18) are correlated to the combustion air flow rate also decreases, reducing the
stoichiometric ratio at the furnace exit, af ; because of the average gas velocity in the furnace as well as the heat
physical dependence of both Cuf and CO on af rather than transfer in the waste heat recovering scheme. As a result, a
on ab : These two stoichiometric ratios are closely related higher exhaust gases temperature is measured at the exit of
throughout the air in-leakage, Da; by the boiler.
ab ¼ af þ Da ð20Þ The three formulae, Eqs. (17), (18) and (22), based on the
expected physical relations among the parameters, were
Air in-leakage, Da; represents the leakage of surrounding
obtained using the STATVIEW commercial code [6], and
air, due to non-air tightness, into the boiler and can be
are valid for steam power values ranging from 20 to 50 t/h
calculated by [5]
and for stoichiometric ratios at the furnace exit, af ; from 1.2
ffiffiffiffiffiffiffiffiffiffiffi
s 
Dnom to 1.6. All the experimental measurements were included in
nom sh
Da ¼ Da ð21Þ the statistical fitting process and the regression coefficients
Dsh
R2 were always greater than 0.9.
where Danom is the air in-leakage at the nominal steam If the set of Eqs. (17), (18) and (22) is introduced into the
power ðDnom
sh ¼ 45 t=hÞ: This parameter, Da
nom
; was methodology to calculate the heat losses q4 ; q3 and q2 ; their
 J. Barroso et al. / Fuel 82 (2003) 1451–1463
Fig. 4. Influence of the steam power and the stoichiometric ratio on the stack gases temperature. Solid lines correspond to Eq. (22) for values of the steam power
of 20 and 50 t/h.
individual contributions to the overall boiler efficiency can
be analyzed. This influence is shown in Fig. 5(a), (b) and (c)
for three different levels of stoichiometric ratio at the exit of
the boiler, ab : 1.45, 1.5 and 1.8, respectively, considering a
fixed constant bagasse moisture of 50% and an ash content of
4% (dry). In all these figures, the stoichiometric ratio at the
furnace exit, af ; has also being included. The evolution of af
as a function of steam power derived from Eqs. (20) and (21)
and its dependence on the level of the stoichiometric ratio at
the boiler exit, ab ; can be easily verified. Comparing the three
figures, it is evidenced that, as the stoichiometric ratio at the
exit of the boiler increases, the heat losses have different
behavior; the exhaust gases heat loss, q2 ; undergoes a
significant rise, while q3 and q4 decrease. Even though Eqs.
(17), (18) and (22) are valid for Dsh of 20 t/h, it can be seen
that for low stoichiometric ratios, Fig. 5(a), ðab ¼ 1:45Þ;
experimental data is only available for steam flows above
30 t/h. During the experimental tests, it was checked that if
the stoichiometric ratio at the furnace exit, af ; is reduced
below 1.2, the boiler starts to work in an unstable regime and,
at the end, combustion stops. In this case, a stoichiometric
ratio at the furnace exit, af ; of 1.2 corresponds to a steam flow
of 28.8 t/h.
On the other hand, as can be observed in this figure, at
higher steam powers, q5 decreases, as predicted by Eq. (14).
As the total heat transfer area is a fixed value (for each
boiler) and the external wall temperature is roughly constant
irrespective of the steam power, then the total heat lost to the
surroundings (in kW) is nearly constant as well. However,
as an increase in the steam power is related to a higher fuel
consumption, a reduction in the conduction heat loss is
finally achieved, according to the behavior also predicted by
Eq. (5).
All these features are summarized in Fig. 6, where the
overall efficiency, h; is plotted as a function of
1459
the stoichiometric ratio at the exit of the boiler and the
steam power, demonstrating the global effect of heat losses
on boiler efficiency (Eq. (1)). Data for only four values of
the stoichiometric ratio are displayed, for clarity. It is
concluded that the highest efficiency is reached for a steam
power value in the vicinity of the nominal one, 45 t/h and for
low values of ab (1.45). This result is supported by the fact
that the largest heat loss in these boilers is that correspond-
ing to the exhaust gases, q2 : However, again for this value of
ab ; a decrease of the steam power below 30 t/h, causes the
unstable combustion regime described before, which finally
results in flame extinction. On the contrary, for ab of 1.5 and
1.6, a nearly flat behavior of the efficiency with respect to
the steam power is reached, for the whole range, with values
quite close to those achieved for the lowest stoichiometric
ratio, ab : It is for this reason that, including in the analysis
the results obtained for all the boilers tested, the optimal
value of the stoichiometric ratio at the exit of the boiler, ab ;
has been determined to range from 1.5 to 1.55, which allows
for a full coverage of the whole range of steam powers. It
should also be noted that, prior to this experimental
research, engineers and boiler operators used to run the
boilers at higher stoichiometric ratio values at the exit of the
boiler, even exceeding 1.8, loosing a large amount of
thermal energy resulting in a lower efficiency.
5.3. Optimization of the heat recovery scheme
As commented in Section 4, and considering the results
obtained in the boiler efficiency analysis in Section 5.2, to
the importance of the exhaust gases heat loss, q2 on the
overall combustion efficiency has been evidenced. In this
section, the optimization of the stack temperature will be
analyzed, for its strong influence on the exhaust gas
enthalpy and, therefore, on q2 :
1460 J. Barroso et al. / Fuel 82 (2003) 1451–1463
Fig. 5. Calculated heat losses ðq2 ; q3 ; q4 ; q5 Þ vs. steam power for three levels
of stoichiometric ratio at the exit of the boiler ðab Þ : (a) 1.45, (b) 1.5 and (c)
1.8. A 50% of bagasse moisture and an (dry) ash content of 4% have been
considered.
Boiler design optimization is a very complex problem
[7], requiring some assumptions to simplify its mathemat-
ical treatment. In this study, to optimize the waste heat
recovery scheme, the speed of the flue gas, steam, water and
air flow are supposed to have their optimal values for all the
heat transfer surfaces studied. At the same time, the furnace
exit gas temperature is kept constant at 900 8C and the steam
power is fixed at the nominal value of 45 t/h. It is also
Fig. 6. Overall boiler efficiency calculated using Eq. (1), and experimental
data vs. steam power for four different stoichiometric ratios at the exit of the
boiler, ab ; with the same conditions considered in Fig. 5.
necessary to consider the specific heat at the exit of the
boiler and its average value at the different heat transfer
surfaces to be independent of the stack temperature. The
heat transfer coefficients should also be considered as
independent of the optimized temperature.
To solve Eqs. (9) and (12), a thermal analysis of all the
heat transfer surfaces in the boiler was first performed
followed by a coupled mathematical and graphical analysis.
The overall heat transfer coefficient ðkÞ of each individual
heat transfer surface considered in the analysis was initially
computed using its own thermal equation. These equations
are slightly different depending on the type of construction,
flow arrangement, core configuration, etc. but they are
always a function of local heat transfer coefficients, g1 and
g2 ; and the thermal efficiency, C; according to the relation
ðg1 g2 Þ
k ðkW=ðm2 KÞÞ ¼ C ð23Þ
ðg1 þ g2 Þ
where subscript 1 means hot gas and 2 refers to the cold fluid
(water, steam or air). A summary of the heat transfer
coefficients for the different surfaces studied is presented in
Table 1. Note that for the bagasse dryer, the volumetric heat
transfer coefficient is given in kW/(m3 K).
The economical cost of the individual heat transfer
surface per area unit was calculated according to the
principle of scaling economy [8] carefully detailed in Ref.
[9], using the equation
 a
P2 F2
¼ ð24Þ
P1 F1
where the same nomenclature as in Eq. (9) has been used. In
this equation, subscript 1 is assigned to the known heat
transfer surface and 2 refers to that to be calculated; a is a
scaling exponent determined with the known cost and heat
transfer area. Results for the individual heat transfer
surfaces considered in this study are depicted in Table 2.
 J. Barroso et al. / Fuel 82 (2003) 1451–1463 1461

Table 1
Heat transfer coefficients used for the different surfaces studied

Generating Air Economizer Superheater Bagasse

tubes heater dryer

k (kW/(m2 K)) 0.041 0.014 0.057 0.054 0.1014a

a
This entry has units of kW/(m3 K).

Table 2
Cost of the individual heat transfer surfaces obtained from the scaling
behavior given by Eq. (24)

Generating Water- Air Economizer Superheater

tubes walls heater

Pi ($/m2) 32 134 16 26 87

For the case of the bagasse dryer, a cost of 300 $/m3 is

obtained. For all the calculations, a recovery coefficient of
Fig. 7. Results obtained for the optimization of the stack temperature for
0.26 yr21 for the furnace, and 0.28 yr21 for the rest of heat case III.
transfer surfaces, has been considered.
A typical value for the stack temperature for boilers that
obtained. In this sense, the optimal heat recovery scheme is
burn common fuel oils and coal ranges between 150 and
case III formed by an economizer followed by an air heater,
300 8C to avoid acid corrosion. However, as stated in
in the exhaust gas flow direction.
Section 2, sulfur contents in bagasse are negligible, yielding
Another important aspect is to ascertain the influence of
a low dew point temperature for the exhaust gases, around
both heat transfer area and conventional fuel costs on the
60 8C, based on experimental determinations. Therefore,
optimized parameters (stack temperature, hot air and
keeping the external pipe temperature in the last equipment
bagasse moisture). To obtain the optimum value as a
over 70 8C (or the stack temperature over 80 8C) the
function of a certain individual cost, a similar procedure to
problem of acid deposition is avoided. Five heat recovery
that described above to optimize the stack gas temperature,
schemes, previously described in Section 4, were con-
has to be followed. For example, to analyze the influence of
sidered, to reduce the stack temperature to its optimal value.
the air heater cost on the optimum value for the stack gas
The non-linear relations given by Eqs. (9), (12) and (24) temperature in case III, the optimum value from the total
were solved for the different waste heat recovery schemes cost chart is obtained for a first air heater cost, then for a
using a computer code based on TKSOLVER software [10] second one and so on, until the range under study is
through the coupled thermal and economical analysis. The completed. In this case, the optimal stack gas temperature is
behavior of the total cost curve against the stack temperature obtained as a function of the air heater heat transfer area cost
is depicted in Fig. 7 for case III. As can be observed, the curve, as depicted in Fig. 8(a). The rest of the heat transfer
optimal stack temperature was found to be 86 8C for this areas and the conventional fuel costs, previously calculated,
particular waste heat recovery scheme, with an optimum are kept constant at the values given in Table 2. In a similar
value of 88 8C for the hot air temperature. way, an analysis was performed to determine the influence
Results for all the heat recovery schemes studied are of the conventional fuel cost on the stack temperature, and
displayed in Table 3. It is observed that the optimal stack the results are shown in Fig. 8(b). As can be observed, if the
temperature, Teg ; varies between 80 and 100 8C for all the conventional fuel cost is increased, the optimum value for
cases analyzed, except when a bagasse dryer is taken into
account (case V). For case V, the optimal stack temperature
Table 3
is slightly higher than 60 8C; which is permissible when the Optimization results for the different combinations studied (see notation in
last recuperative piece of equipment is the bagasse dryer, text)
because acid deposition on the previous heat transfer surface
Combination Teg TAH W Efficiency, h Cost, Z
(in this case the economizer) will never occur. The optimal (8C) (8C) (%) (%) ($/yr)
value for bagasse moisture, W; is near 41% when a bagasse
dryer is taken into account (case V). The optimal hot air Case I 83.6 150.0 50.0 87.9 709,436
temperature, TAH ; reaches the highest value for case II, as Case II 98.6 257.0 50.0 86.8 716,418
expected. Finally, if the thermal results presented in Table 3 Case III 86.0 88.0 50.0 87.8 700,206
Case IV 98.0 25.0 50.0 86.9 701,385
are coupled with the total cost displayed in the last column,
Case V 62.6 145.0 41.4 90.5 816,553
the most efficient combination of heat surfaces can be
1462 J. Barroso et al. / Fuel 82 (2003) 1451–1463
Fig. 8. Individual dependence of the optimum stack temperature with
respect to (a) heat transfer surface area cost and (b) conventional fuel cost
for the economizer–air heater heat recovery scheme (case III).
the stack temperature decreases. This behavior can be
explained because the reduction in the consumption of
conventional fuel becomes more significant if its price is
raised, and, therefore, the optimum stack temperature is
lower. On the other hand, if the cost of air heater heat
transfer area increases, the optimum value for the stack
temperature also increases, as depicted in Fig. 8(a), because
a further increment in the heat transfer area to reduce the
stack temperature becomes more expensive. Hence, the
optimum stack temperature displaces to lower values, if
either a reduction in the cost of air heater heat transfer area
or an increase in the conventional fuel cost, occurs.
6. Conclusions
A methodology for the determination of heat losses and
efficiency of RETAL boilers has been established, and a
simplified Bagasse-boiler Industrial Test Code has also
been implemented for the evaluation of boilers using
bagasse as fuel. The use of this methodology requires a
10 h test to evaluate and optimize the operation of a boiler,
saving a lot of time when compared with the nearly 81 h
needed when the ASME or GOST methodologies are
applied.
Three statistically-based models, Eqs. (17), (18) and
(22), have been obtained, which are used to calculate the
fixed carbon, chemical carbon, and exhaust gas heat losses,
as a function of the stoichiometric ratio at the furnace, af ;
and the steam power, Dsh : The individual contributions of
all heat losses on the overall boiler efficiency have been
determined, yielding an efficiency chart for these bagasse-
boilers and the range of optimal stoichiometric ratio at the
exit of the boiler, ab : For all the boilers analyzed, the
optimal value of this parameter ranges between 1.5 and
1.55, for the whole range of steam power, contrasting with
the 1.8 typically used prior to this study. This result greatly
simplifies the decision-making and workload of the
engineers and boiler operators, who can now easily control
the boiler for optimum performance in different operating
regimes.
To optimize the retrofit waste heat recovery scheme,
several computer codes have been developed. The optimal
stack temperature lies in the range between 60 and 100 8C,
much lower than the 200 8C commonly measured in most of
the boilers installed in the Cuban sugar cane industry.
According to a coupled thermal and economic analysis, the
optimal heat recovery scheme is that formed by an
economizer followed by an air heater, in the direction of
the exhaust gas flow, giving an optimal stack temperature of
86 8C, with a hot air temperature of 88 8C.
Acknowledgements
This research has been partially funded by the EDIMEC
Company through Project PCT-013. Several members of the
University of Matanzas contributed to this work. Their
collaboration is gratefully acknowledged. Authors are
indebted to Dr Peter Walsh for the useful technical
discussion and his many suggestions to the original
manuscript.
References
[1] ASME CODE and STANDARDS PTC 4.1 Performance test code
steam generating units, USA; 1972 (date of specific issue adopted:
1964).
 J. Barroso et al. / Fuel 82 (2003) 1451–1463
[2] Triembomblia V. Thermotechnical test in boilers. Moscow: MIR;
1979, in Russian.
[3] Hugot E. Handbook for sugar engineers, 3rd ed. S.A. Mexico: CECSA
Continental; 1974, in Spanish.
[4] Bagasse-boiler Industrial Test Code. Universidad de Matanzas; 1996,
in Spanish
[5] Rubio A. Steam boilers. La Havana: MES; 1982, in Spanish.
[6] SAS Institute, Software STATVIEW for Windows Version 5.01.
USA: SAS Institute; 1998.
1463
[7] Andriuchenko A. Thermal cycles and process optimization in the
electric power station. Moscow: MIR; 1974, in Russian.
[8] Portuondo F. Industrial enterprise economy. La Habana, Cuba: Pueblo
y Educación; 1983, in Spanish.
[9] Barroso J, Amaveda H, Ochoa M. Bagasse boiler optimization. Centro
Azúcar 1999; in press, in Spanish.
[10] Universal Technical Systems, Software TKSOLVER Ver-
sion 3.32. USA: Universal Technical Systems; 2000. http://www.
uts.com.